Only V / Only J alors qu'il n'y a que du bruit
Le mail que j'ai envoyé à McGill m'a mis la puce à l'oreille : vraiment, leurs données RNA-Seq contiendraient des dizaines de % de reads en only V ou only J ? rbx : ~/patients/u061-McGill/V03-372-1T_1.fq
Regardé quelques reads au début (-i -x 1000 -K -uU )... et non, on fait n'importe quoi :
Bla CTAGGCATGGCTCCTCTCCACAGGAAAACTCCACTCCAGTGCTCAGCTTGCACCCTGGCACAGGCCAGCAGTTGCTGGAAGTCAGACACCTGAGAAGAAC
1 ! seed IGK UNSEG only V _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _-K _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Je pensais qu'il y avait un filtre de e-value pour only V / only J...
Différence entre 2015.10.2 et 5298c86f, sur V03-372-1T_1.fq (http://rbx.vidjil.org/browser/?patient=1233&config=25, 10M reads) : UNSEG strand -> 52957 99.5
- UNSEG too few V/J -> 7341235 97.5
- UNSEG only V -> 1660661 94.3
- UNSEG only J -> 1041862 94.7
- UNSEG ambiguous -> 898556 99.4
- UNSEG too few V/J -> 10786960 96.9
- UNSEG only V -> 108604 98.5
- UNSEG only J -> 46512 99.5
- UNSEG ambiguous -> 238 100.0 UNSEG too short w -> 1 100.0
Cette correction de bug peut justifier 2015.12 :-)